Migrating Large System Prompts to Ollama Without Attention Loss

A practical engineering guide to moving complex cloud LLM system prompts to local Ollama runtimes with RoPE frequency scaling and Modelfile tuning.